Landscape Dynamics Determine the Small-Scale Genetic Structure of An

Journal of Coastal Research 28 4 780–786 West Palm Beach, Florida July 2012

Landscape Dynamics Determine the Small-Scale Genetic Structure of an Endangered Dune Slack Plant Species Dries Bonte{, Peter Breyne{, Rein Brys{{, Eduardo de la Pen˜a{, Bram D’hondt{, Ce´line Ghyselen{, Martijn L. Vandegehuchte{, and Maurice Hoffmann{{ www.cerf-jcr.org

{Ghent University {Research Institute for Nature and Forest Department of Biology Department of Biodiversity and Natural Environment Terrestrial Ecology Unit Kliniekstraat 25 K.L. Ledeganckstraat 35 1070 Brussel, Belgium 9000 Ghent, Belgium [email protected]

ABSTRACT

Bonte, D.; Breyne, P.; Brys, R.; de la Pen˜ a, E.; D’hondt, B.; Ghyselen, C.; Vandegehuchte, M.L., and Hoffmann, M., 2012. Landscape dynamics determine the small-scale genetic structure of an endangered dune slack plant species. Journal of Coastal Research, 28(4), 780–786. West Palm Beach (Florida), ISSN 0749-0208.

Understanding the processes that determine genetic variation within landscapes is a crucial factor for successful management of threatened plant species that are sensitive to both environmental and genetic bottlenecks. While current insights point to the importance of historical landscape processes for the genetic structure of populations at large spatial scales, their relevance at small spatial scales has been largely neglected. In this context, coastal dunes are a typical example of dynamic and geologically young landscapes in which current and historical sand drift may have strong impacts on the spatial dynamics of a large number of plant species. One of these is the endangered plant species Parnassia palustris, typically inhabiting dune slacks formed by recent sand displacements in parabolic dune landscapes. Dune slacks originating from the same sand drift process are located within the same parabola unit. The species is known to suffer from dispersal limitation and from inbreeding when genetic exchange between populations is restricted. By means of amplified fragment length polymorphism, we demonstrate that the species shows a genetic substructuring both at the level of the metapopulation and the local landscape. Populations located within the same parabola unit are much more closely related than expected on the basis of geographic distance. Moreover, population size is related to genetic diversity within populations. The species’ population genetic structure should consequently be regarded as a shifting mosaic of genetic variation, mediated by sand drift driven landscape formation. Therefore, the maintenance of sand dynamics is essential to preserve genetic diversity in dynamic dune landscapes.

ADDITIONAL INDEX WORDS: AFLP, metapopulation, Parnassia palustris, population size, sand dynamics.

INTRODUCTION biology (Manel et al., 2003). Recent population genetic studies have demonstrated the relationship between population size, Understanding the processes and patterns of gene flow and distance between populations, and genetic differentiation local adaptation requires a detailed knowledge of how among populations, which is strongly affected by the species’ landscape characteristics affect the genetic structure of set of life history traits (see, e.g., Hamrick and Godt, 1996; populations (Holderegger and Wagner, 2008). This is especially Honnay and Jacquemyn 2007; Nybom and Bartish, 2000 for true for plants, in which the process of gene flow is often comprehensive reviews). In contrast, only few studies have strongly influenced by the configuration of the surrounding explicitly tested how landscape genesis (i.e., processes that landscape and prevented by the barriers within it (Sork et al., shape the landscape configuration) may affect spatial patterns 1999). This understanding is crucial, not only to improve in neutral genetic variation, underlying microevolutionary general insights on genetics in conservation biology, but also to processes, and the eventual feedback on life history traits properly manage endangered species in order to maintain and (Balkenhol et al., 2009; Storfer et al., 2007). restore the genetic diversity of their populations. Approaches In general, studies aiming to test the relative importance of from so-called landscape genetics (Holderegger and Wagner historicalvs.currentgeneflowinrelationtolandscapegenesisare 2008) promise to facilitate our understanding of how geograph- constrained by the fact that the configuration of most landscapes ic and environmental features may affect genetic variation at to date is heavily influenced by anthropogenic factors on a very both the population and individual level and may have short timescale relative to most natural processes of landscape important implications for ecology, evolution, and conservation formation. Therefore, the observed genetic variation of these populations often does not confirm general predictions of (meta-) population genetic theory, for instance, the fact that genetic DOI: 10.2112/JCOASTRES-D-10-00128.1 received 24 August 2010; accepted in revision 1 November 2010. distance and geographic distance are not always found to be ’ Coastal Education & Research Foundation 2012 related (Bonnin et al., 2002; Leimu et al., 2006). Sand Dynamics and Plant Genetic Structure 781

Landscape formation may induce spatial genetic variation ulation crosses resulting in a higher reproduction capacity than due to historic gene flow and common founder effects and by within-metapopulation crosses, indicated a larger degree of decreasing local population sizes if habitat formation becomes compatibility between plants originating from different meta- restricted. Two consequences of small population size are populations (Bossuyt 2007). By means of amplified fragment increased genetic drift and inbreeding. Genetic drift and length polymorphism (AFLP) analysis, we tested the hypoth- inbreeding may influence small plant populations by changing esis that landscape dynamics (i.e., parabolic dune genesis) are patterns of genetic diversity. Genetic drift decreases genetic important drivers of genetic differentiation in P. palustris variation within populations, while it generally increases within existing metapopulations relative to among-metapopu- differentiation among populations, an effect that becomes lation differentiation. more pronounced in declining populations (Barrett and Kohn, 1991; Ellstrand and Elam, 1993; Young, Boyle, and Brown, MATERIAL AND METHODS 1996). Inbreeding increases homozygosity within populations, whereas smaller populations lose genetic variation faster than Study Species larger populations (Ellstrand and Elam, 1993). The expected Parnassia palustris (Saxifragaceae) is a perennial herb with relationship between population size and genetic diversity is, a circumboreal distribution (Bonnin et al., 2002; Borgen and however, often missing (Bonnin et al., 2002; Leimu et al., 2006) Hultga˚rd, 2003). Each plant has basal leaves and 1–30 because populations occur in human managed habitats that flowering stems, each with one terminal flower. The species could have experienced dramatic short-term changes. As is hermaphroditic and protandrous. The five stamens sequen- Bonnin et al. (2002) suggest, these changes may include tially discharge their pollen before the stigma becomes bottlenecks and a changing degree of connection to other receptive. It is usually cross-pollinated and rarely autoga- populations (Holderegger and Wagner, 2008). mously pollinated (Martens, 1936). The species is mainly In dynamic landscapes, landscape genesis is by definition a outcrossing and strongly depends on pollinators for optimal geologically young process that can often be disentangled from pollination and seed set (Sandvik and Totland, 2003). Parnas- human interference on habitat configuration. Coastal dunes sia palustris is insect pollinated, mainly by Diptera, particu- along the North Sea comprise such a system. Owing to sand larly hoverflies (Syrphidae), but other insects contribute to drift dynamics, dry parabolic dune ridges are continuously pollen transfer (Sandvik and Totland, 2003). It flowers from shifting in a northeasterly direction, leaving behind young August through September, and fruits, containing several ephemeral dune slacks (Provoost, Jones, and Edmondson, hundreds of seeds, ripen from September through October. The 2009), whose ecological characteristics are often determined by small, light seeds are dispersed by water and wind. Parnassia the groundwater level and the carbonate content at the palustris is considered endangered in northern France, moment of development. Here, several plant species find their Luxembourg, Belgium, and the Netherlands (Bonnin et al., optimal, though temporary, growing conditions; they are often 2002), where it is a rare plant in lime-rich dune slacks along the restricted to specific environmental conditions in space and coast and inland lime-rich marshes. time. Their populations are subsequently expected to be predominantly linked to the historical dynamics of these Study Area landscapes, in which gene flow patterns are strongly determined by the occurrence of dry parabolic dune ridges that may The study was conducted in a sandy dune area along the function as effective dispersal barriers. On the other hand, the western Belgian and northeastern French coast. This area has fragmentation of larger dune entities due to the steady increase been extensively fragmented by human activities since the of urbanized areas is another factor that may induce genetic beginning of the 20th century. The remaining dune areas are variation at larger geographic scales. So, the historical intact protected as nature reserves and managed by mowing and and connected coastal dune region has been fragmented by extensive grazing. Within these landscapes, dune slacks are urbanization into several entities, further referred to as the frequently formed by wind erosion down to the level of the metapopulations. Separated populations occur in dune slacks, groundwater at the backside of moving dune ridges, termed which are due to common history situated in larger parabola parabolic dunes. Such dune slacks, with a calcareous and units. Populations are subsequently hierarchically located in nutrient-poor soil, flood in winter, dry up in summer, and form parabola units situated within one metapopulation (i.e., the a temporary natural habitat for P. palustris. Parnassia dune entity). palustris populations are consequently restricted to individual Bonnin et al. (2002) analyzed the population genetic parabolic units, each of these representing separate landscape structure of 14 Parnassia palustris populations in northern entities with a common history of origin. Populations of P. France. They compared populations occurring in different palustris only persist in the relatively early-successional stages habitats and found restricted gene flow among populations. of vegetation establishment within these dune slacks. Further However, panmixia within eight of the populations suggested succession leads to dune shrub formations. However, the that, in contrast to seed dispersal, pollen transfer was not mowing of shrub encroached dune slacks may also result in limited at a regional scale. In contrast to the populations the potential persistence of P. palustris populations. Both studied by Bonnin et al. (2002), the studied P. palustris natural parabola dynamics and human conservation actions populations in this work are closely located in isolated may thus determine the spatial structure of the species in the metapopulations within a small region along the Flemish– remaining intact dune areas (further referred to as metapop- French coast. A pollination experiment, with outside-metapopulations; see Bossuyt 2007; Bossuyt and Honnay 2006;

Journal of Coastal Research, Vol. 28, No. 4, 2012 782 Bonte et al.

Table 1. Genetic diversity and population characteristics of the studied P. palustris populations (n, number of sampled plants for population genetic

analysis; PPL, percentage of polymorphic loci; Hj, expected genetic diversity). The symbols in the population ID refer to the samples metapopulation: P, Perroquet; T, Ter Yde; W, Westhoek.

Population Patch Area 2 Population n PPL Hj Size (m ) P1 19 83.80 0.32 200 2808 P2 18 85.30 0.34 180 2466 P3 18 79.40 0.31 120 2342 P4 17 82.40 0.32 100 723 T1 19 88.20 0.33 50,000 4105 T2 19 82.40 0.32 300 2506 T3 10 82.40 0.32 125 2596 T4 20 86.80 0.34 220 11,070 T5 20 86.80 0.35 1400 4798 T6 7 80.90 0.31 14 413 T7 10 79.40 0.29 10 234 T8 10 82.40 0.30 27 263 W1 18 85.30 0.35 2400 9064 W2 17 89.70 0.32 350 10,227 Figure 1. Location of the sampled P. palustris populations along the coast W3 19 86.80 0.31 40,000 5196 of Flanders (B) and Northern France (Fr). P, Perroquet populations; W, W4 16 85.30 0.32 3200 960 Westhoek populations; T, Ter Yde populations. Light blue lines indicate W5 17 83.80 0.33 700 3910 parabolic dunes, shifting to the northeast, leaving behind dune slacks on W6 18 88.20 0.34 2400 23,953 the weather side. In blue: remaining dune areas along the coast. W7 19 91.20 0.35 125 1547 W8 9 83.80 0.32 9 821

Bossuyt, Honnay, and Hermy, 2003). Parnassia palustris consequently experiences two levels of fragmentation along this coast: large-scale fragmentation due to urbanization, and was performed in two steps using the primer combinations small-scale fragmentation due to the combined action of EcoRI+A/MseI+C and EcoRI+C/MseI+G for preamplification parabolic dune dynamics, shrub encroachment, and mowing and EcoRI+ATC/MseI+CAT, EcoRI+ACA/MseI+CAC, and management. EcoRI+CCA/MseI+GTT for selective amplification. Fragment Based on the study of Bossuyt (2007), we selected 20 separation and detection took place on a Nen IR2 DNA analyzer populations of P. palustris within three metapopulations (Licor) using 24 cm denaturing gels with 6.5% polyacrylamide. (Figure 1). Each population is thus restricted to one dune IRDye size standards (50 to 700 bp) were included for sizing of slack. In the Ter Yde metapopulation (T, 260 ha) we sampled the fragments. Control samples were included in each gel to eight populations, ranging in size from three to more than check for reproducibility within and between gels. Only clear, 10,000 individuals. All plants were randomly selected in the intense polymorphic bands between 75 bp and 500 bp were population, so both at the periphery and the center. This scored. Scoring was done using the SAGAmx software (Licor). metapopulation is separated from the two other metapopula- We scored the presence or absence of every marker in each tions, Westhoek (W, 340 ha) and Perroquet (P, 225 ha), by a individual as 1 or 0 (present or absent) to form a binary data densely urbanized area of 10 km where no individuals of P. matrix. palustris occur. The metapopulations W and P are separated by a road and a camping site at the border between Belgium and Genetic Data Analysis France. We sampled eight and four populations in the W and P metapopulations, respectively, ranging in size between seven Based on the allele frequencies, within-population and and more than 10,000 individuals. An overview of the sampled metapopulation genetic diversity was estimated by the per- populations is given in Table 1. centage of polymorphic loci (PPL) and Nei’s genetic diversity (expected heterozygosity, Hj). Additionally, we determined the AFLP Analysis proportion of total genetic variability within a population compared to the total genetic variability recorded among the

Leaves sampled in the field were immediately frozen in three metapopulations (population differentiation, FST) and liquid nitrogen. In the lab, they were freeze-dried for 48 hours total metapopulation diversity (Ht; Lynch and Milligan, 1994). and homogenized with a mill (Retsch MM 200) to fine powder. The number of permutations for the test on FST was 999. These We used 20 mg of dried leaf material for DNA extraction using measures were calculated using AFLP-Surv (Vekemans et al., the QuickPickTM plant DNA kit (Isogen Life Science). DNA 2002). To assess the degree of molecular variation within and quality was checked on 1.5% agarose gels. Concentration and among populations, total genetic diversity was partitioned by purity were determined using a ND-1000 spectrophotometer applying a hierarchical analysis of molecular variance (NanoDrop, Thermo Scientific). We used 100 ng of DNA for (AMOVA; Table 2) on Euclidean pairwise genetic distances AFLP analysis according to Vos et al. (1995). Restriction and using Genalex 6.1 (Peakall and Smouse, 2006). Significances ligation were done in a single step. Amplification of fragments were determined based on 999 permutations. The ØST is an

Journal of Coastal Research, Vol. 28, No. 4, 2012 Sand Dynamics and Plant Genetic Structure 783

Table 2. Hierarchical partitioning of the genetic variation among metapopulations (MP). For each source of the variation (within populations, among populations from the metapopulation, and among metapopulations), we provide the statistical attributes from the MANOVA (df, degrees of freedom; SS,sumof squares; MS, mean squares; Est. Var., estimated variance component; and variance explained). The different variance contributions are denoted by specific genetic statistics (Stat) that are tested for their significance level (Value, Prob).

Source df SS MS Est. Var. Variance Explained Stat Value Prob

Among MP 2 375.191 187.595 1.471 13% ØRT 0.125 0.001 Among pops/MP 17 440.178 25.893 0.997 8% ØPR 0.097 0.001 Within pops 317 2946.807 9.296 9.296 79% ØPT 0.210 0.001 Total 336 3762.175 222.784 11.763

analogue for FST values used for dominant markers such as 0.364, while the overall level of genetic differentiation (Fst)was AFLP and was derived from the Euclidean genetic distances. Its 0.11 6 0.08 and highly significant, indicating substantial significance was calculated using the Monte Carlo procedure in genetic substructure. This is confirmed by partitioning (Table 2) Genalex 6.1 (999 permutations). Pairwise genetic distances of the total genetic variation among the metapopulations among the populations and their level of significance were (MANOVA, ØRT 5 0.125, P , 0.001). Genetic variation within obtained from AMOVA. Again 999 permutations were applied. each of the three metapopulations was lower, but substantial

The relationship between pairwise genetic distances (FST), and significant (among-population variation, MANOVA, ØPR 5 derived from AFLP-Surv and geographic distances (Table 3), 0.097, P , 0.001), while the largest amount of variation was due was assessed with Mantel tests implemented within the ade4 to individual variation within each population (within-popula- package of R statistical software v.2.6.0 (R Development Core tion variation, MANOVA, ØPT 5 0.210, P , 0.001). Genetic team; 999 replicates). Similarly, differences in pairwise genetic diversity within populations subsequently accounted for nearly distances (FST) between populations from the same or different 80% of the observed genetic variation. Variation among and parabola entities were tested by a distance-based nested within metapopulations was 13% and 8%, respectively. When permutational analysis of variance (PERMANOVA). considering all populations, genetic differentiation is driven by isolation by distance (Mantel test, r 5 0.632, P , 0.001). RESULTS Population Genetic Structure Within-Metapopulation Genetic Diversity

The three AFLP primer combinations rendered 68 highly Within metapopulations, genetic differentiation is apparent- reliable polymorphic markers. Genetic diversity within popu- ly not driven by isolation by distance (Mantel test for Ter Yde lations was high, with the percentage of polymorphic loci (PPL) [T], r 5 0.0147, P 5 0.434; Westhoek [W], r 5 20.413, P 5 ranging from 79.4 to 91.2. Concordantly, the expected 0.991; Perroquet [P], r 5 0.223, P 5 0.421). For all three heterozygosity (Hj) ranged from 0.29 to 0.35. There were no metapopulations, the partitioning of the genetic diversity is obvious differences in genetic diversity levels between the largest at the within-population level (MANOVA; Table 3). For three metapopulations. Total genetic diversity (Ht) reached W, T, and P, 9%, 6%, and 13% of the diversity can be attributed

Table 3. Hierarchical partitioning of the genetic variation within each of the metapopulations, so disentangling within populations, among populations from each parabola unit, and among parabola units from the specific metapopulation (Westhoek, Ter Yde, Perroquet). We provide the statistical attributesofthe MANOVA (df, degrees of freedom; SS, sum of squares; MS, mean squares; Est. Var., estimated variance component, and variance explained). The different variance contributions are denoted by specific genetic statistics (Stat) that are tested for their significance level (Value, Prob). Partitioning of the genetic diversity within each of the three metapopulations.

Source df SS MS Est. Var. Variance Explained Stat Value Prob Westhoek (W)

Among parabolas 2 56.810 28.405 0.034 0% ØRT 0.003 0.233 Among pops/parabola 5 132.533 26.507 0.969 10% ØPR 0.092 0.001 Within pops 134 1274.279 9.510 9.510 90% ØPT 0.095 0.001 Total 141 1463.622 64.421 10.512 Ter Yde (T)

Among parabolas 1 27.728 27.728 0.156 2% ØRT 0.016 0.001 Among pops/parabola 6 108.049 18.008 0.608 6% ØPR 0.062 0.001 Within pops 111 1025.156 9.236 9.236 92% ØPT 0.076 0.001 Total 118 1160.934 54.972 9.999 Perroquet (P)

Among parabolas 1 46.403 46.403 0.364 3% ØRT 0.034 0.004 Among pops/parabola 2 68.655 34.327 1.358 13% ØPR 0.131 0.001 Within pops 72 647.371 8.991 8.991 84% ØPT 0.161 0.001 Total 75 762.429 89.721

Journal of Coastal Research, Vol. 28, No. 4, 2012 784 Bonte et al.

isolation by distance at larger geographical scales. At smaller spatial scales, genetic substructuring according to the landscape unit (i.e., the parabola) rather than by geographic distance was observed. The loss of genetic variation in small populations and drift induced additional differentiation among populations. The species’ level of variation is in line with studies on species with similar life history at a similar spatial scale (Hamrick and Godt, 1996; Honnay and Jacquemyn 2007; Nybom and Bartish, 2000). However, since genetic differentiation increases with geographic distance (Nybom and Bartish,

2000), our reported Fst values are lower than those recorded in studies at larger regional or even continental scales (see for instance Bonnin et al., 2002). Our results point to the importance of considering landscape genesis in order to understand historical and contemporary gene flow patterns in plant populations. Although this is indeed previously acknowledged for plants by the seminal paper of Sork et al. (1999), empirical studies at this small scale in Figure 2. Relationship between expected heterozygosity within each of terrestrial systems are scarce (but see Honnay et al., 2009). the sampled P. palustris populations and population size, based on linear regression (see text). Some studies explicitly testing how river bank geometry and dynamics impact the genetic structure in plants (Honnay et al., 2009; Kondo, Nakagoshi, and Isagi, 2009; Kudoh and Whig- to the population level within parabola unit, respectively. ham, 1997; Van Looy et al., 2009) and arthropods (Lambeets, Among these parabola units, however, variation between Breyne, and Bonte, 2010) showed that ‘‘isolation by functional populations is low or nonexistent. (dis)connectivity’’ rather than ‘‘isolation by distance’’ is a more The genetic substructuring is, nevertheless, largely deter- important driver for genetic diversity. Our study subsequently mined by landscape formation, with populations within the acknowledges this approach, even at small spatial scales. same parabola units showing higher levels of genetic similarity AFLP analysis revealed isolation by distance when consid- compared with those from different units. This is reflected by ering the three metapopulations, or the combined Westhoek– the outcome of the PERMANOVA, in which significantly Perroquet metapopulation, but not when contrasting genetic different pairwise Fst values between dune areas were found with geographic distance within each of the single metapopu- (pseudo F 5 11.116, P 5 0.025) and between parabola units lations. These observations are in contrast with the study of within the three different metapopulations (pseudo F 5 2.221, Bonnin et al. (2002), who did not find any sign of isolation by P 5 0.038). distance even at larger spatial scales by using allozyme or In addition to the mentioned differentiation between parabola cpDNA markers. While the different genetic techniques may be units, within-metapopulation differentiation may also be driven responsible for the contradicting patterns observed, the overall by drift and the loss of genetic variation in small populations. landscape context of the study is probably a more tentative This is evidenced by the positive relationship between within- alternative explanation. Bonnin et al. (2002) studied 14 French population genetic diversity and population size as revealed by populations in different contrasting habitat types with a long history of fragmentation due to landscape formation at large mixed model regression (b 5 0.008 6 0.003 SE, F1,17 5 4.63, P 5 0.047) and correlation (r 5 0.34, P , 0.05; see Figure 2) and the geological scales (chalk hills, marshes, old and young dune fact that the number of pairwise differentiations with other systems). In contrast, our study emphasized patterns of genetic populations from the same metapopulation decreases with diversity within a young coastal dune landscape that only population size (r 5 20.680 6 0.255 SE, Wald x2 5 7.19, df 5 1, experienced fragmentation due to urbanization after the P 5 0.0073), when corrected for differences in average genetic Second World War (Provoost et al., 2004). The combined action differentiation between the three metapopulations (df 5 2, Wald of genetic drift, natural selection, and limited gene flow x2 5 12.06, P 5 0.0024). No differences in genetic diversity between distantly isolated P. palustris populations is therefore within populations were found between metapopulations (effect suggested to destroy any signature of isolation by distance in of metapopulation and metapopulation 3 log [population size], Bonnin’s French populations, while this may not be the case in our study area. This is additionally confirmed by the higher both F2,13 , 2.18, P . 0.05). Because population sizes differ levels of genetic differentiation in the studied populations of neither between parabola units (F5,11 5 1.17, P 5 0.343) nor Bonnin et al. (2002), both with isozyme and cpDNA markers. between the different metapopulations (F2,11 5 1.90, P 5 0.195), differences in population size do not underlie the retrieved The genetic substructuring within each of the metapopula- differentiation between parabola units. tions, according to the landscape history (and parabola formation), suggests that gene flow is restricted among DISCUSSION parabola units. It is, however, remarkable that nearby populations of different parabola units, even at the opposite This study clearly indicated that the genetic structure of our side of a parabola dune unit, are genetically more different than studied coastal dune populations of P. palustris is shaped by distant populations within the same parabola unit. These dune

Journal of Coastal Research, Vol. 28, No. 4, 2012 Sand Dynamics and Plant Genetic Structure 785 ridges often reach up to more than 10 m above the slacks and ACKNOWLEDGMENTS are partly vegetated with shrubs. This suggests that these parabola dunes may act as efficient gene flow barriers, either We thank Bea Bossuyt for help with the sampling of for pollen or seeds. The tiny seeds of P. palustris are well Parnassia material, thus providing a solid basis for this ´ adapted to wind or water dispersal (Bouman et al., 2000; paper. Bram D’Hondt, Celine Ghyselen, and Martijn Vande- Sandvik and Totland 2003). Because coastal dune slacks gehuchte are Ph.D. students funded by the Research inundate during winter, dispersal by water during the winter Foundation—Flanders (FWO). Eduardo de la Pen˜ a and Rein inundation may be responsible for gene flow among some Brys are postdoctoral fellows funded by the FWO. Dries Bonte populations from the same parabola unit and constrain seed is supported by the EVE-NET FWO-research network. dispersal among parabola units that are separated by higher dune ridges, acting as watersheds. LITERATURE CITED Although similar within-parabola dispersal may be expected Balkenhol, N.; Gugerli, F.; Cushman, S.A.; Waits, L.P.; Coulon, A.; for wind dispersal, it is less obvious for pollen transfer. Arntzen, J.W.; Holderegger, R., and Wagner H.H., 2009. Identify- According to Ennos (1994), pollen migration contributes 20 ing future research needs in landscape genetics: where to from here? Landscape Ecology, 24, 455–463. timesmoretogeneflowamongpopulationsthanseed Barrett, S.C.H. and Kohn, J., 1991. The genetic and evolutionary migration in P. palustris (if in mutation-drift equilibrium). consequences of small population size in plant: implications for Personal observations of one of us indicated that P. palustris is conservation. In: Falk, D. and Holsinger, K.E., (eds.), Genetics and predominantly pollinated by the hover fly Episyrphus balteatus Conservation of Rare Plants. Oxford, U.K.: Oxford University (Degeer). In a study in cultivated landscapes, Wratten et al. Press, pp. 3–30. Bonnin, I.; Colas, B.; Bacles, C.; Holl, A.C.; Hendoux, F.; Destine, B., (2003) found field boundaries to be barriers for pollen transfer and Viard, F., 2002. Population structure of an endangered species with a serious decline over distances of 200 m, suggesting that living in contrasted habitats: Parnassia palustris (Saxifragaceae). even common pollinators like E. balteatus may have a limited Molecular Ecology, 11, 979–990. impact on pollen transfer in structured landscapes with Bonte, D.; Maelfait, J.-P., and Lens, L., 2006. Sand dynamics in pronounced geomorphological barriers. coastal dune landscapes constrain diversity and life-history characteristics of spiders. Journal of Applied Ecology, 43, 735–747. Because P. palustris has been shown to be sensitive to the Borgen,L.andHultga˚rd, U.M., 2003. Parnassia palustris: a loss of genetic variation within and among metapopulations genetically diverse species in Scandinavia. Botanical Journal of (Bossuyt 2007), resulting in a fitness decline due to the partial the Linnean Society, 142, 347–372. self-incompatibility, conservation of genetic diversity and gene Bossuyt, B., 2007. Genetic rescue in an isolated metapopulation of a naturally fragmented plant species, Parnassia palustris. Conser- flow within and among metapopulations appears to be vation Biology, 21, 832–841. essential for the maintenance of population viability in the Bossuyt, B. and Honnay, O., 2006. Interactions between plant life long term. Bossuyt (2007) found no evidence for outbreeding span, seed dispersal capacity and fecundity determine metapopu- depression on seed set when crossing plants from different lation viability in a dynamic landscape. Landscape Ecology, 21, metapopulations, but, of course, it could manifest after several 1195–1205. Bossuyt, B.; Honnay, O., and Hermy, M., 2003. An island biogeo- generations. Our study showed that the population structure graphical view of the successional pathway in wet dune slacks. should be regarded as a shifting mosaic of genetic variation, Journal of Vegetation Science, 14, 781–788. affected by sand drift–mediated landscape formation. Out- Bouman, F.; Boesewinkel, D.; Bregman, R.; Devente, N., and crossing with genetically different pollen enhances fitness Oostermeijer, G., 2000. Verspreiding van zaden. Utrecht, The Netherlands: Stichting Uitgeverij van de Koninklijke Nederlandse (Bossuyt 2007). The conservation of this within-metapopula- Natuurhistorische Vereniging. 240p. tion genetic variation due to the maintenance of natural sand Ellstrand, N.C. and Elam, D.R., 1993. Population genetic consequenc- drift, in combination with some gene flow among parabola es of small population-size—implications for plant conservation. dunes, is therefore likely to be an important attribute to Annual Review of Ecology and Systematics, 24, 217–242. maintain the species’ fitness at the highest level. Bossuyt Ennos, R.A. 1994. Estimating the relative rates of pollen and seed migration among plant populations. Heredity, 72, 250–259. (2007) showed that especially small populations benefited from Hamrick, J.L. and Godt, M.J.W., 1996. Effects of life history traits on outcrossing with genetically different pollen. Consequently, in genetic diversity in plant species. Philosophical Transactions of the addition to restoring large population sizes, P. palustris may Royal Society of London, Biological Sciences, 351, 1291–1298. particularly benefit from genetic rescue. An increased genetic Heidinger, I.; Hein, S., and Bonte, D., 2010. Patch connectivity and variation within both local populations and metapopulations is sand dynamics affect dispersal-related morphology of the blue- winged grasshopper Oedipoda caerulescens in coastal grey dunes. therefore very likely to be advantageous for population Insect Conservation and Diversity, 3, 205–212. persistence, especially because high estimates of genetic Holderegger, R. and Wagner, H.H., 2008. Landscape genetics. differentiation suggest that gene flow among metapopulations Bioscience, 58, 199–207. is disrupted as a result of high levels of urbanization. This Honnay, O. and Jacquemyn, H., 2007. Susceptibility of common and study therefore adds evidence on the importance of natural rare plant species to genetic consequences of habitat fragmentation. Conservation Biology, 21, 823–831. sand dynamics for the conservation of genetic variation within Honnay, O.; Jacquemyn, H.; Van Looy, K.; Vandepitte, K., and metapopulations in addition to the positive effects of natural Breyne, P., 2009. Temporal and spatial genetic variation in a sand dynamics for the conservation of biodiversity at the metapopulation of the annual Erysimum cheiranthoides on stony species level (Bonte, Maelfait, and Lens, 2006; Maes and Bonte river banks. Journal of Ecology, 97, 131–141. Kondo, T.; Nakagoshi, N., and Isagi, Y., 2009. Shaping of genetic 2006) and at the level of ecological functioning (Bonte, Maelfait, structure along Pleistocene and modern river systems in the and Lens, 2006; Heidinger, Hein, and Bonte, 2010; Vandege- hydrochorous riparian Azalea, Rhododendron ripense (Ericaceae). huchte, de la Pen˜ a, and Bonte, 2010). American Journal of Botany, 96, 1532–1543.

Journal of Coastal Research, Vol. 28, No. 4, 2012 786 Bonte et al.

Kudoh, H. and Whigham, D.F., 1997. Microgeographic genetic review. Journal of Coastal Conservation. DOI: 10.1007/s11852-009- structure and gene flow in Hibiscus moscheutos (Malvaceae) 0068-5 populations. American Journal of Botany, 84, 1285–1293. Sandvik, S.M. and Totland, O., 2003. Quantitative importance of Lambeets, K.; Breyne, P., and Bonte, D., 2010. Spatial genetic staminodes for female reproductive success in Parnassia palustris variation of a riparian wolf spider Pardosa agricola (Thorell, 1856) under contrasting environmental conditions. Canadian Journal of on lowland river banks: the importance of functional connectivity in Botany—Revue Canadienne De Botanique, 81, 49–56. linear spatial systems. Biological Conservation, 143, 660–668. Sork, V.L.; Nason, J.; Campbell, D.R., and Fernandez, J.F., 1999. Leimu, R.; Mutikainen, P.; Koricheva, J., and Fischer M., 2006. How Landscape approaches to historical and contemporary gene flow in general are positive relationships between plant population size, plants. Trends in Ecology and Evolution, 14, 219–224. fitness and genetic variation? Journal of Ecology, 94, 942–952. Storfer, A.; Murphy, M.A.; Evans, J.S.; Goldberg, C.S.; Robinson, S.; Lynch, M. and Milligan, B.G., 1994. Analysis of population genetic Spear, S.F.; Dezzani, R.; Delmelle, E.; Vierling, L., and Waits, L.P., structure with RAPD markers. Molecular Ecology, 3, 91–99. 2007. Putting the ‘‘landscape’’ in landscape genetics. Heredity, 98, Maes, D. and Bonte, D., 2006. Using distribution patterns of five 128–142. threatened invertebrates in a highly fragmented dune landscape to Vandegehuchte, M.L.; de la Pen˜ a, E., and Bonte, D., 2010. Relative develop a multispecies conservation approach. Biological Conser- importance of biotic and abiotic soil components to plant growth and insect herbivore population dynamics. PloS One, 5, e11174. vation, 133, 490–499. Van Looy, K.; Jacquemyn, H.; Breyne, P., and Honnay, O., 2009. Manel, S.; Schwartz, M.K.; Luikart, G., and Taberlet, P. 2003. Effects of flood events on the genetic structure of riparian Landscape genetics: combining landscape ecology and population populations of the grassland plant Origanum vulgare. Biological genetics. Trends in Ecology and Evolution, 18, 189–197. Conservation, 142, 870–878. Martens, P. 1936. Pollination et biologie florare chez Parnassia Vekemans, X.; Beauwens, T.; Lemaire, M., and Roldan-Ruiz, I., 2002. palustris L. Bulletin of the Societ’e Royale Botanique de Belgique, Data from amplified fragment length polymorphism (AFLP) 68, 183–231. markers show indication of size homoplasy and of a relationship Nybom, H. and Bartish, I.V., 2000. Effects of life history traits and between degree of homoplasy and fragment size. Molecular Ecology, sampling strategies on genetic diversity estimates obtained with 11, 139–151. RAPD markers in plants. Perspectives in Plant Ecology, Evolution, Vos, P.; Hogers, R.; Bleeker, M.; Reijans, M.; van de Lee, T.; Hornes, and Systematics, 3: 93–114. M.; Frijters, A.; Pot, J.; Peleman, J.; Kuiper, M., and Zabeau M., Peakall, R. and Smouse, P.E., 2006. GENALEX 6: genetic analysis in 1995. AFLP: a new technique for DNA fingerprinting. Nucleic Acids Excel. Population genetic software for teaching and research. Research, 23(21), 4407–4414. Molecular Ecology Notes, 6, 288–295. Wratten, S.D.; Bowie, M.H.; Hickman, J.M.; Evans, A.M.; Sedcole, Provoost, S.; Hoffmann, M.; Bonte, D., and Leten, M., 2004. J.R., and Tylianakis, J.M., 2003. Field boundaries as barriers to Landschap en beheer van kustduinen. In: Hermy, M.; De Blust, movement of hover flies (Diptera : Syrphidae) in cultivated land. G., and Slootmakers, M. (eds.), Natuurbeheer. Leuven: Davidsfonds Oecologia, 134, 605–611. i.s.m. Argus, Natuurpunt en Instituut voor Natuurbehoud. Young, A.; Boyle, T., and Brown, T. 1996. The population genetic Provoost, S.; Jones, M.L., and Edmondson, S.E. 2009. Changes in consequences of habitat fragmentation for plants. Trends in landscape and vegetation of coastal dunes in northwest Europe: a Ecology and Evolution, 11, 413–418.

Journal of Coastal Research, Vol. 28, No. 4, 2012